Anyway, for reference for anyone looking into this in the future. Google Cloud Data Fusion is almost there, you can use the JDBC connector by setting up a new Connection via Database (NOT Oracle/anything else). BUUT, you can't actually use it in the "Replicate" accelerator to load only changed rows. You could implement a CDC solution in a JAR file and use a custom plugin, but the further away I can be from Java/Oracle the happier I am 🤢 Another option is to build a custom program/script/whatever, that connects via ODBC (because ew Java), loads the data, and then pushes it to BigQuery, and running it in a Cloud Function for example (or CKE, Cloud Run, my grandma's computer, etc.)

extracting data might be also possible using SuiteScript but considering the load, it might be better done outside.